We sought to determine how knowledge and experience affect performance during clinical simulations. METHODS: Three groups varying in levels of experience, knowledge, and skill in respiratory care (5 students, 5 therapists with no more than 2 years of experience, and 5 instructors with 7 years or more of experience, 'experts') completed a computerized simulation in respiratory care. The simulation involved care of a cardiac arrest patient in a coronary care unit, with mechanical ventilation and weaning complications. The program kept track of positive decisions (correct responses), negative decisions (errors), and the quality and amount of information requested. In addition, all subjects were asked to 'think aloud' regarding their deliberations during the section on initiation of mechanical ventilation. RESULTS: Statistical analyses indicated that students commit more information-gathering errors than do experts and ask for more information and experts make more efficient use of the information that they request. Analyses of thinking-aloud protocols indicate that the experts' comprehensive knowledge-base enables them to make ventilator-setting decisions based on the patient's clinical picture, whereas students rely more on instrumentation. CONCLUSION: Our findings help explain the differences between expert and novice performance but do not provide information that can be directly applied to the instructional process. Continued study of expert performance may allow the development of advice-strategies that reduce the amount of experience necessary to acquire higher levels of competence and allow novices to acquire the necessary knowledge in less time. (Respir Care 1988;33:779-785.)